EP0064131B1 - Digital modulation method - Google Patents

Abstract

In a multi-phase modulation method in which digital words each having a word length of at least 3 bits control the generation of a modulated carrier in a manner such that each digital word value is associated with a particular amplitude and phase of the modulated carrier, and in which the modulated carrier is formed of a combination of three oscillating signals having identical frequencies, shifted in phase by 120 DEG relative to one another, and each having a selected amplitude, the digital words are associated with the oscillating signals in such a manner that a first group of digital words is represented only by modulated carrier phases which possess equal angular spacings in the 360 DEG phase plane and a plurality of further digital words is represented by a zero amplitude carrier formed by combining all three oscillating signals each having the identical selected amplitude or by suppressing all three oscillating signals.

Description

The present invention relates to a digital modulation method using digital signals with a k bit word length (k ₌ 1, 2, 3 ...), according to which the 2 ^k possible bit words are assigned modulation signals with specific amplitudes and phase positions, these modulation signals being a combination of three Carrier oscillations of constant frequency, phase-shifted by 120 ° with respect to one another and whose amplitude is weighted, are used as modulation signals, only those whose vectors divide the 360 ° phase plane into equally large angular sections, and in addition a level-less zero signal is used as the modulation signal.

Such a method is known from IEEE Transactions on Communications, Vol. COM-21, No. 10, October 1973, pages 1108-1115.

From DE-PS 2 206 382 a digital modulation method is known in which digital signals, especially those whose bit word lengths are greater than two, are used to modulate a carrier signal. The aim here is to modulate the carrier signal so that it is as insensitive as possible to interference on the transmission link in order to be able to recover the original data signal from the modulated carrier signal without errors in the receiver. With word lengths of the digital signal of 2 and perhaps also 3 bits, it is sufficient to modulate the carrier signal with a constant amplitude only in its phase. Limits are set to pure phase modulation very quickly, since as the number of possible bit words increases, the phase difference between the signals assigned to these bit words decreases. If the phase differences are too small, there is a slight risk that a signal will assume the phase of its phase-adjacent signal even with smaller phase fluctuations. Therefore, in order to be able to implement modulation that is as insensitive to interference as possible with higher-quality data signals, the phase-amplitude modulation is switched over, i. that is, the signals associated with the possible bit words differ in phase and / or in amplitude. The choice of the phase and amplitude differences between the individual signals should now be made so that common disturbances occurring during signal transmission, such as noise, phase shifts, sloping amplitudes, multipath propagation and the like, do not disturb the signal amplitudes and phases so strongly that transitions between different ones Signal states occur. The document DE-PS 2 206 382 mentioned at the outset specifies a modulation method based on the phase amplitude modulation, which has a greater signal-to-noise ratio than conventional methods (eg quadrature amplitude modulation QAM). In contrast to the QAM, in which the different signal states are created by the combination of two carrier signals controlled by the bit words and phase-shifted by 90 ° with respect to one another and whose amplitude is weighted accordingly, three methods are phase-shifted and amplitude-weighted by 120 ° against one another according to the method described in DE-PS2206382 Carrier vibrations combined with each other; namely after. According to this method, the signal vectors are chosen with respect to their amplitude and phase so that their ends lie in the centers of equally sized regular hexagons, which are arranged in a honeycomb-like manner around the vector origin M in the phase plane (see FIG. 1).

The phase plane shown in FIG. 1 shows that the angles between some signal vectors, e.g. B. between the vectors 1 and 11 or 5 and 15 etc., are quite small, whereby transitions can easily take place between these vectors due to phase shifts. With regard to phase shifts, this modulation method is therefore still quite susceptible to interference. In addition, the non-uniform angular division of the phase plane by the vectors is unfavorable for demodulation, since the demodulator must be able to precisely recognize several very different phase states. If, as here, the clear determination of the phase states is made more difficult, the probability of error in the demodulation increases.

An improvement in the probability of error of the method described in DE-PS 2 206 382 is brought about by the introduction outlined in IEEE Transaction's on Communications, Vol. COM-21, No. 10, October 1973, pages 1108-1115 known digital modulation methods. According to this method, the modulation signals are also formed by combining three carrier oscillations which are 120 ° out of phase with one another and whose amplitude is weighted. The fact that only those are used as modulation signals whose vectors divide the 360 ° phase plane into equally large angular sections, and in addition a level-less zero signal is used as the modulation signal, to which a bit word is assigned, the distances between the individual vectors increase Modulation signals and thus their susceptibility to interference.

The invention is based on the object of specifying a digital modulation method of the type mentioned at the outset which, compared to the prior art, brings a further improvement in terms of susceptibility to interference and which ensures that the modulated signals are recognized as error-free as possible.

This object is achieved according to the invention in that when modulating data signals whose word length is greater than two bits, the zero signal is assigned a plurality of bit words which differ in at least three bit positions and in the period in which a bit word is modulated in series, together with the for this modulation signal belonging to this bit word, two further redundant modulation signals are transmitted, which are formed from two bit words that are derived from the true bit word and coded in such a way that the true bit word can be correctly determined during demodulation on the redundant modulation signals.

Appropriate embodiments of the method according to the invention are evident from the subclaims.

The fact that the signal with the level zero is assigned to one or even better even several bit words lowers its average power compared to the modulation signal generated by the method mentioned at the beginning, provided that all signal states are transmitted with the same frequency. This proves to be particularly advantageous when there is a demand for the lowest possible average signal power.

• Figur 1 eine Signalvektorverteilung in der Phasenebene für 16 Bitworte nach dem Stand der Technik,
• Figur 2 eine Signalvektorverteilung in der Phasenebene für 4 Bitworte,
• Figur 3 eine für 16 Bitworte,
• Figur4 eine Signalvektorverteilung für 8 Bitworte, wobei dem Null-Vektor zwei verschiedene Bitworte zugeordnet sind und
• Figur für diesen Fall ein Diagramm, nach der Erfindung, welches die Zuordnung, der Bitworte zu den Modulationssignalen verdeutlicht.

The invention will now be explained in more detail with reference to exemplary embodiments shown in the drawing. Show it :

• FIG. 1 shows a signal vector distribution in the phase plane for 16 bit words according to the prior art,
• FIG. 2 shows a signal vector distribution in the phase plane for 4 bit words,
• 3 shows a for 16 bit words,
• FIG. 4 shows a signal vector distribution for 8 bit words, two different bit words being assigned to the zero vector and
• Figure in this case a diagram, according to the invention, which illustrates the assignment of the bit words to the modulation signals.

1 has already been pointed out in connection with the description of the prior art. The regular hexagons of the same size drawn in there in the phase plane represent an aid for the construction of the signal vectors. If the vectors 1 to 16 end in the centers of these hexagons, there are quite favorable signal-to-noise ratios between the vectors.

2 shows a very simple vector system for a data signal with four 2-bit words; namely, according to the invention, a bit word is assigned signal 0, which has no level and of course no defined phase state, and signals are assigned to the remaining three bit words, whose vectors 1, 2 and 3 have the same amplitudes and phase differences of 120 °. So-called interference circles are drawn in around the vectors. They delimit the range in which an interference voltage vector can rotate around the tips of the signal vectors 0, 1, 2, 3, without one signal state being falsified by another due to the interference of the interference voltage. This means that in the event of a signal distortion caused by Noise, phase shifts, sloping amplitudes or multi-path reception in the radio field, the demodulator then correctly recognizes the original signal state as long as the vector tip does not move out of the associated interference circuit. The further apart the interference circles are, the less likely it is that one signal state will be falsified into another.

Since the zero signal is also used in the system in FIG. 2, which is generated by superimposing the three carrier oscillations of the same amplitude, which are phase-shifted by 120 °, or by transmitting none of the three carrier oscillations, the other three signals can be very far in phase ( namely 120 ° apart, which results in an optimal signal-to-noise ratio between the signals. The signal vectors 1, 2 and 3 shown in FIG. 2 have the same length. It is also possible to form three differently long vectors from the combination of differently weighted carrier vibrations. This is useful in the extreme case of the 3-phase system, when the phase difference of 120 ° is not sufficient to clearly recognize the signal states in the receiver in the case of very strong phase disturbances. The stepped signal amplitudes then offer another criterion for distinguishing the signals.

3 shows a vector system for a data signal with sixteen 4-bit words. Here too the zero signal 0 is used. However, only one bit word is assigned to the zero signal in this exemplary embodiment. There is an ideal distribution of the signal vectors in the phase plane if, as in the known modulation method according to FIG. 1, each vector end is placed in the center of a regular hexagon. The other hexagons are clustered around a central hexagon, the center of which coincides with the vector of the zero signal, which is also the origin point 0 of vectors 1 to 15.

This hexagonal configuration means that the associated vectors divide the 360 ° phase plane into equal angular sections. Each vector therefore forms an angle of 30 ° with its two neighboring vectors. Modulation signals are obtained which can be demodulated with a very low probability of error due to optimally large and equidistant phase spacings; Should the amplitude spacing between the signals not be sufficient in the case of the densest hexagon arrangement shown in FIG. 3 in the case of very strong disturbances, the hexagons 8, 9, 10, 11, 12, 13, 14 and 15 of the outer hexagon ring can be further away from the origin point 0 become what an extension of the vectors ending in the centers of these hexagons.

The hexagon system intended as a construction aid can, as the dotted hexagons in FIG. 3 indicate, be enlarged for further signal states.

To the demodulator on a fixed reference Synchronize phase and also set the gain on the demodulator side can be superimposed on a modulation signal of a defined phase in pulses at certain time intervals. In the exemplary embodiment, this is the modulation signal 8. The associated vector is briefly extended as a result, which is indicated by dashed lines in FIG. 3.

Using a simple exemplary embodiment for three-valued bit words of a digital signal, it will now be shown that only one bit word can be assigned to the zero signal. The consequence of this is that fewer signal states with an amplitude other than zero are necessary in order to represent all bit words. The result of this is that fewer vectors have to be distributed in the 360 ° phase plane. The phase distances between the vectors can therefore be increased further. The advantage of the largest possible phase spacing has already been mentioned above.

4 shows the phase level with the vectors of the modulation signals 1 to 8 for digital signals with a 3-bit word length. 4 shows the assignment of the 8 possible bit words to the modulation signals 1 to 8. There are 6 vectors in the phase plane, namely those of the modulation signals 1 to 6, which each have a phase distance of 60 ° with respect to their neighboring vectors. The remaining two bit words 000 and 111 are assigned to the zero signal. In order to be able to recognize during demodulation which of the two bit words has been assigned to a transmitted zero signal, additional information must be transmitted from which it can be derived whether bit word 000 or bit word 111 has been transmitted by the zero signal.

The additional information provides two redundant bit words, which are also modulated and transmitted together with each modulated bit word of the data signal to be transmitted.

The formation of suitable redundant bit words is illustrated in the pulse diagram in FIG. 5. T denotes the clock signal for the modulator. Any data signal DS to be modulated has been selected, the three-value bit words of which are 101, 011, 000, 001 and 110. The designations a, b and c stand for the three carrier oscillations which are phase-shifted by 120 ° relative to one another, and the combination of which enables the desired modulation signals to be generated. If a 1 is present at a, b or c, this means that the associated carrier oscillation is also used to form the modulation signal. If a 0 is present at a, b or c, the associated carrier oscillation is not involved in the resulting modulation signal. The data signal is fed to the modulator in series. The first bit of a word determines whether the carrier oscillation a (with phase 0 °), the second bit whether the carrier oscillation b (with phase 120 °), and the third bit whether the carrier oscillation c (with phase 240 °) contributes to the modulation. The digital signal DS is fed serially to the modulator, i. that is, sizes a, b and c are set one after the other. The value table for a, b and c in the lower part of FIG. 5 illustrates this process (see framed bit values of a, b and c). The first bit of a word brings a to the value specified by the bit, the second bit sets b to it and finally the third bit sets c to the corresponding bit value. At the end of a data signal word, a, b and c have the states 1 or 0 corresponding to the word, so that by combining the three carrier oscillations in accordance with the states of, a, b and c, the modulation signal associated with the data signal word is produced.

It is common practice to transmit only the final modulation signal associated with the data signal word. This can then have a pulse length that corresponds to the time it takes for a data signal word connected serially to the modulator to set a, b and c to the correct states. However, this period should now be used not only for the transmission of one modulation signal but two further redundant modulation signals. The two redundant modulation signals are based on two bit words, the first of which differs from the subsequent second, and the second differs from the following true data signal word at most in one bit position. This is achieved by maintaining the states of a, b and c between two changes caused by the data signal. Compare the table of values for a, b and c in the lower part of FIG. 5. With the aid of these two redundant bit words transmitted as modulation signals, it can be determined in the receiver whether bit word 000 or bit word 111 in the modulator is transmitted to a zero signal has been assigned. If the previous redundant bit word was 000, 100, 010 or 001, the zero signal was based on bit word 000, because from these three redundant bit words only one word position can result from changing one bit position. In the same way, bit word 111 has certainly been assigned to the zero signal if the previous redundant bit word was 111, 011, 101 or 110. Instead of the previous redundant bit words, the subsequent redundant bit words can also be used to assess the zero signal. In addition, the two redundant bit words are used for error detection in all other transmitted data signals. Under certain circumstances, error correction is also possible using the redundant bit words.

It is also possible to carry out another more complex coding for the two redundant bit words, which permits very reliable error detection and correction.

Since according to the above method in the same period, which is usually only for the transmission of the one assigned to the data signal Modulation signal is available, two additional modulation signals must now be transmitted, the three modulation signals are to be coordinated in a very specific way. Namely, the three modulation signals are transmitted with a 120 ° phase shift with respect to their envelope. Only the sine or cosine fundamental wave, which is obtained by suppressing all harmonics of the otherwise rectangular envelope, serves as the envelope of each modulation signal. According to this method, advantageously no greater bandwidth of the transmission channel is required than for the transmission of a modulation signal with a rectangular envelope, the width of which corresponds to the period for the serial delivery of a data signal word to the modulator.

The method described in the exemplary embodiment for 2 ³ = 8 bit words can of course also be transferred to digital signals with more significant bit words, e.g. For example, with 2 ⁴ = 16 bit words, four different bit words can already be assigned to the zero signal. It should only be bit words that differ from one another in at least three bit positions.

Claims (4)

1. Digital modulation method with the aid of digital signals of a word length of k bits (k = 1, 2, 3 ...), according to which modulation signals of certain amplitudes and phase portions are associated with the 2^k possible bit words, wherein these modulation signals are formed by the combination of three carrier oscillations of constant frequency, which are displaced in phase through 120° one relative to the other and weighted in their amplitude, wherein only such are used as modulation signals, the vectors of which divide up to 360° phase plane into equally great angular portions, and a level-free zero signal is additionally used as modulation signal, characterised thereby, that several bit words differing in at least three bit places are associated with the zero signal in the modulation of data signals, the word length of which is greater than two bits and that in the time span, in which a bit word is modulated serially, two further redundant modulation signals are transmitted together with the modulation signal belonging to this bit word and are formed out of two bit words which are derived from the true bit word and so coded that it can be determined free of error from the redundant modulation signals during the demodulation, which true bit word has been transmitted.

2. Method according to claim 1, characterised thereby, that the bit word 000, which is formed by emission of one of the three carrier oscillations displaced through 120°, and the bit word 111, which is formed by combination of all three carrier oscillations at equal amplitude, are associated with the zero signal in the modulation of trivalent bit words.

3. Method according to claim 1, characterised thereby, that the envelopes of both the redundant modulation signals and of the modulation signal associated with the true bit word are displaced in phase through 120° each relative to the other.

4. Method according to claim 3, characterised thereby, that the envelopes are sine waves or cosine waves.

Images